Film vs Digital - Dynamic Range

[mauro_franic]

At second examination, it does not appear to be the glass texture. It appears to be noise introduce with the GLASS+ICE combination. Still this may be too high frequency to have a direct impact on the print.

[svenja_mathews]

@HC Lim

 

You already said it in your answer. Given you have a stellar optics, shooting into the sun is perhaps easier with film...unless you do HDR (but hey, carrying around this tripod all the time).

 

S.

[hclim]

I also shoot towards the sun with lots of people moving about. How is that done in HDR?

 

Come to think of it, nowadays people don't talk much about lens flare. It must be pretty obvious why.

[mauro_franic]

I verified indeed the fine noise introduced with ICE+Glass Holder+Sharpening does not show on the prints. Perceptually, it may even have a smoothing effect on very large prints.

 

I tested on my Epson 3800 at: 500dpi (18x22 print), 360dpi (25x31 print), and 200dpi (45x55 print).

[tim_klimowicz]

Wow.

[svenja_mathews]

Hi HCLIM,

 

With film, it is easy. With HDR, you can merge manually the layers with different exposure. By now, the only way to get the job done. Automatic mergers usually produce ghosts. But perhaps at some time in the future this issue becomes resolved.

 

Lens flare? Maybe this is no issue with the kit lenses? I'm not sure, though...

 

Regards - S.

[peter_ferling]

You don't need the darkroom. I just have my negs processed by a local shop that I trust. I get a good whiff of

the developer while there to keep me fixed... :)

 

Then I just scan MF negs on an epson flatbed, and get 30" prints.

 

For 35mm, hate to admit, but I secure those in the neg holders that came with my scanner, put them on a light

table and shoot them as raw with a 40D or 1Ds and a 100mm F/2.8 macro (sharp sharp lens that one). Reverse and

recover contrast in Lightroom using a preset on batch capture and edit right along with my digital shots. 24

exposures takes only twenty minutes.

 

Now whenever I encounter a scene that requires the DR of film. I bag the digital and shoot with my T70 or FTBn.

In fact, I'll experiment with the digital first (pixels are free and at least I get the shot), and follow up

with film. I use a lot less film that way, and have more keepers.

[chris_werner]

Double Wow.

 

How can anyone follow a thread this long?

[rishij]

Jack:<i>"Film has log color but it's still scanned by a linear scanner. There's no such thing as film. Everything is scanned & viewed on a computer eventually, using the exact same linear photodiode technology."</i>

<p>

Right, but I hope you're not trying to say then that the final product of film + scan gives you a linear recording of the original scene. Because it doesn't. Film (negative) compresses the tonal range of the original scene; the scan then makes a linear representation of that compressed tonal range (which it shouldn't have any trouble doing; that is, it should be able to recover the entire recorded range on the film itself with 14 bit A/D conversion saved to 16 bit files), and so you end up with a file with a compressed tonal range of the original scene. You then expand the tonal range carefully using whatever adjustments you want to bring about your final vision of the scene.

<p>

The digital camera just gives you a linear representation of the original scene. OK it's not that simple b/c the RAW converter does a lot, and there's a lot more processing involved (including, these days, 'highlight tone priority', but the effects of that are debatable), but you get the picture.

<p>

Important point: <b>film + scanning</b> does NOT give you a linear representation of the original scene, which a digital camera otherwise does.

<p>

Rishi

[rishij]

Chris,

<p>

<i>"How can anyone follow a thread this long?"</i>

<p>

Clearly you've not seen this thread:<br>

<a href="http://www.photo.net/film-and-processing-forum/00ROOo">Wow - read this re: Film versus Digital debate!</a>

<p>

Cheers,<br>

Rishi

[rishij]

Kelly:<i>"One can have a poster shot with a cellphone and folks will like it if you say its film; and hate it if

yoiu say it was shot with a cellphone; "since a cellphone is never a camera; and never will be" according to many

folks. Its actually an interesting exercise in human bias.</i>

<p>

Great point. It's called 'synthesizing happiness'. Check out the incredible TED talk given by Dan Gilbert

entitled "Why are we happy?":

<p>

<a href="

Gilbert asks: Why Are We

Happy?</a>

<p>

Rishi

[rishij]

Mauro,

 

I don't really see the texture of the (presumably anti-newton) glass in the sharpened image -- could you point it out?

 

Unless you're talking about the fine grain color noise across the entire image... which I assumed was just due to the sharpening.

 

I have seen texture of AN glass on a LS-4000 scan... but it's quite big and not fine grained. This is AN glass from focal point.

 

I have been able to subdue it, however, with an optical diffuser.

 

I need to write a thread on it :)

 

Rishi

[vijay_nebhrajani]

Umm, I don't get the whole log vs linear thing - it is dynamic range that matters, so lets say a dynamic range of

80 db (10,000:1) is a dynamic range of 80 dB = 13 plus stops = 10,000:1 = 4.0 and so on. Shapes of curves don't

really matter, you can always convert between curve shapes. If it is log, apply an antilog function. If it is

linear and you want log, just take the logarithm etc.

<p>

Film may well have 80 dB dynamic range, but digital sensors will eventually get there if they aren't already

there. Beyond that, someone needs to explain why digital giving a "linear" representation matters at all.

<p>

And please explain what you mean by film having a non-linear response. It is linear - a look at any film

datasheet will confirm this. For <a

href="http://www.kodak.com/global/en/professional/support/techPubs/e4046/e4046.pdf">the new Ektar 100</a> for

example, you'll see curves that show density vs log-exposure. Density is itself a logarithm, so the response is

exactly linear. What am I missing here?

<p>

If you mean the S-shaped curve with a heel and a toe - that is just how real linear systems behave: they have

clipping points; that is not non-linearity. Between the heel and toe there is a linear region; this is how even

electronic sensors work.

[mauro_franic]

Rishi, yes I am referring to the noise introduces with the Glass+ICE+Sharpen combination. It so happens this noise is ultra high frequency (very fine) and does not show on the prints.

 

Here is with even more exaggerated sharpening:

 

http://shutterclick.smugmug.com/gallery/6616619_YJEwK#427295802_gSvPJ-O-LB

 

Compare the noise on frames 2 and 3 (Glass+ICE+Sharpen ) with 5 and 6 (No Glass+ICE+Sharpen ).

[mauro_franic]

Vijay, film gets more resistant to light the more exposed it gets. I an over simplified example, it is like walking into a wall,

 

With film the closer you are to the wall the shorter the steps you take (yes, you can always linearize later).

 

With digital (today) you take equally spaced steps and you hit the wall sooner (no, you can't remove the bump on your head by taking smaller steps later when you walk to bed).

[mauro_franic]

Rishi, to clarify, you are correct it is not glass texture it is fine noise. This actually benefits the prints (per sizes I listed above).

[mauro_franic]

And this is my (subjective) reading of the provia test chart resolution captured by the Coolscan 9000 (along the transversal higher resolving axle) :

 

Glass + No ICE 3900

 

Glass + ICE Normal 3750

 

Glass + ICE Fine 3400

 

No Glass + No ICE 3900

 

No Glass + ICE Normal 3850

 

No Glass + ICE Fine 3400

[mauro_franic]

dpi lines per picture height

Glass + No ICE 3900 8598

Glass + ICE Normal 3750 8268

Glass + ICE Fine 3400 7496

No Glass + No ICE 3900 8598

No Glass + ICE N 3850 8488

No Glass + ICE Fine 3400 7496

[mauro_franic]

Not sure what happened there:

 

In DPI:

 

Glass + No ICE: 3900.

 

Glass + ICE Normal: 3750.

 

Glass + ICE Fine: 3400.

 

No Glass + No ICE: 3900.

 

No Glass + ICE N: 3850.

 

No Glass + ICE Fine: 3400.

 

 

_.

 

In lines per picture height:

 

Glass + No ICE: 8598.

 

Glass + ICE Normal: 8268.

 

Glass + ICE Fine: 7496.

 

No Glass + No ICE: 8598.

 

No Glass + ICE N: 8488.

 

No Glass + ICE Fine: 7496.

[brad_farwell]

Glass + ICE + Whiskey = 8432

 

Glass + No ICE + Whiskey = 8200

[brad_farwell]

(No Glass + ICE + Whiskey = Mess)

[mauro_franic]

Very nice!

[rishij]

Vijay, sure there's a region of linear response (sometimes where the slope is higher than 1) -- this is the most 'usable' region, hence you want to place the main subject exposure somewhere in that region. But on the upper end, it becomes harder & harder to expose the film due to the build up of metallic silver at sensitivity sites as well as the building negative charge at sensitivity specks (electron traps).

 

Digital sensors don't do this, AFAIK, because I'm not all-knowing. No one is. Except for Feynman.

 

But thanks for bringing it up... Because I've always wondered if it's right to call it a 'log' response. Also, this all makes sense for negative film, but positive film also works with silver halide emulsions, correct? The difference being that color dye does NOT form where sufficient exposure takes place. So if exposure is capped off in positive film, as it is in negative film, doesn't that mean that highlights would be saved in slide film, because it becomes harder and harder to expose, so there'll still be some chances of color dye FORMING in highlights?

 

I don't get it... please someone explain if you know. Although, sorry Mauro, I should post a separate topic for this question :)

 

Rishi

[rishij]

Mauro -- weird! Why would glass introduce fine color noise?!

<p>

Here's what I saw with AN glass between the LED source and the film in my LS-4000 scanner (w/ no optical diffuser, which the LS-9000 has, keep in mind):

<p>

<img src="http://staff.washington.edu/rjsanyal/FH3_ANglass-Comparison1.jpg" width=800>

<br>

<a href="http://staff.washington.edu/rjsanyal/FH3_ANglass-Comparison1.jpg">Link to full-size image</a>

<p>

It's not a fine texture at all. It's the etching in the glass. What you see is rather strange I have to say... but much easier to get rid of than what I see with my AN glass (AFAIK, impossible to get rid of)... your noise I'm pretty sure we could get rid of using Neat Image.

<p>

And while we're on the subject: isn't Neat Image amazing?? I've tried Noise Ninja, but for film scans, have been very unhappy. What sets Neat Image apart is the fact that you can control how much of high frequency vs mid frequency vs low frequency noise reduction you want... which allows you to single out film grain from the rest of the image details.

<p>

Rishi

[mauro_franic]

Neat image is indeed very nice.

 

The glass surface you posted looks pretty bad/horrible. The Nikon glass holder does not have that problem.

 

The noise that you see on the Glass+ICE example (shows also without the glass holder but not as much) actually has a smoothing effect on the print. I believe if it is an educated intended result from the ICE algorithm.

 

Give it a try and print my test strips on your printer at 300, 360 or 600 or 720 dpi (or any dpi you use on medium format). PLEASE LET ME KNOW YOUR OBSERVATION FROM YOUR PRINTS. MY PRINTS DO NOT SHOW THAT NOISE AT ALL.

 

Use the one with sharpening: "With fine sharpening: http://shutterclick.smugmug.com/gallery/6616619_YJEwK#426487722_6mhBs-O-LB"